Abstract: Recognition of palm-dorsa vein patterns is a new biometric identification technology.The paper presents a new algorithm for extracting such patterns. In the algorithm watershed transformation is firstly used to extract Feature Points of Vein Pattern （FPVP） from the image of palm-dorsa vein pattern. Feature Points of Vein Pattern are then upgraded into Dominant Points with moment filter and count filter for different feature information of FPVP.Every Dominant Point is a multidimensional vector of and the 2-norms of all DPs in an image constitute the feature vector of the vein pattern.In the end the feasibility of extracting feature of palm-dorsa vein pattern is evaluated by performing the matching test of samples with a correlation method.The dataset is of 265 samples from 53 vein patterns,and the minimum verification error rate is about 4.31%.